metalink上的不错的总结
Introduction
An ORA-3113 "end of file on communication channel" error is a general error usually reported by a client process connected to an Oracle database. The error basically means 'I cannot communicate with the Oracle shadow process'. As it is such a general error more information must be collected to help determine what has happened - this error by itself does not indicate the cause of the problem. For example, ORA-3113 could be signalled for any of these scenarios:
- Server machine crashed
- Your server process was killed at O/S level
- Network problems
- Oracle internal errors / aborts on the server
- Client incorrectly handling multiple connections
- etc.. etc.. etc.. - a lot of possible causes !!
This short article describes what information to collect for an ORA-3113 error. It is common for this error to be accompanied by other errors such as:
- ORA-1041 internal error. hostdef extension doesn't exist
- ORA-3114 not connected to ORACLE
- ORA-1012 not logged on
These are all symptomatic of being disconnected from Oracle.
Please collect as much information as possible from the items listed below and submit this information to Oracle support. Where a step produces an output file / trace file this may be needed by Oracle Support to help determine the cause of the error. The more information you can present in one go the better your chance of a speedy solution. Note that some sections may not be applicable.
If viewing this article on a Web Browser you can follow the links, otherwise manually go to the relevant section.
What Scenario does the ORA-3113 occur in ?
A. When attempting to startup Oracle ? -> Section A B. When attempting to make a connection to the database ? -> Section B C. Client gets the error running SQL / PLSQL ? -> Section C D. Server trace file reports ORA-3113 ? -> Section D You may find it useful to scan the checklist in Section E at the end of this article. This covers some questions / issues relevant to all problem sections.
(A) ORA-3113 when attempting to STARTUP Oracle
There are several phases involved in starting up a database. If ORA-3113 occurs during startup then abort the instance and start up using the sequence below. If an error occurs at any step then see the related notes below. a. Start any required services. On error see A1 Eg: On NT start the OracleServiceSID b. Connect as a SYSDBA user. On error see A1 Eg: svrmgrl SVRMGR> connect / as sysdba (or connect internal) c. Startup nomount. On error see A1 Eg: SVRMGR> startup nomount d. Mount the database. On error see A1 and A2 Eg: SVRMGR> alter database mount; e. Recover the database On error see A3 Eg: SVRMGR> recover database f. Open the database On error see A4 Eg: SVRMGR> alter database open; g. Wait 3 minutes then issue a select. On error see A4 Eg: SVRMGR> select count(*) from DBA_OBJECTS; A1) Errors connecting as SYSDBA / Internal OR on startup nomount There is something fundamental wrong with the software / environment if you cannot connect to Server Manager as a DBA user. The steps here cover errors such as ORA-3113, ORA-12547: TNS:lost contact or similar errors connecting to Oracle or starting the instance NOMOUNT. Check the following items: A1.1) If possible reboot the server disabling any automatic startup of Oracle before you do so. This may seem drastic but helps make sure you are working from a consistent starting point. A1.2) Check your environment points at the expected ORACLE_HOME and ORACLE_SID and that TWO_TASK is not set (Unix) or LOCAL is not set (NT registry). Check the USER_DUMP_DEST and BACKGROUND_DUMP_DEST and default trace directories under this environment for any user trace files or alert log entries generated. These may help indicate the cause of the problem. Eg: ORA-600[SKGMINVALID] may indicate a problem with the shared memory Unix parameters on Unix systems. Try to show that any trace file / alert log entry you find is truely related to the "CONNECT" command by re-issuing the "connect" and checking for a new trace file / alert entry at the time of the error. A1.3) Unix only: Some Unix platforms need LD_LIBRARY_PATH to be set correctly to resolve any dynamically linked libraries. As the user with the problem: % script /tmp/ldd.out % id % cd $ORACLE_HOME/bin % ldd oracle % exit If the 'ldd' command does not exist go to the next step below. Check that all lines listed show a full library file. If there are any 'not found' lines reported contact Oracle support with the output of /tmp/ldd.out . A1.4) Unix only: Your 'oracle' executable may be corrupt. Relink it thus: Log in as the 'oracle' user. % script /tmp/relink.out % cd $ORACLE_HOME/rdbms/lib % mv $ORACLE_HOME/bin/oracle $ORACLE_HOME/bin/oracle.dd.mon.yy % rm -f ./oracle % make -f ins_rdbms.mk ioracle % exit Prior to Oracle7.3 the relink command was: make -f oracle.mk ioracle If this reports any errors Oracle support will need to see the contents of the file /tmp/relink.out . A1.5) Have you installed the Parallel Server Option ? ORA-3113 can occur if you have installed the Parallel Server Option but do NOT have a Distributed Lock Manager installed or running correctly. Unix: If the Parallel Server Option was installed by accident then it can be de-installed by relinking. Eg: Shut down any Oracle instances % script /tmp/relink.out % cd $ORACLE_HOME/rdbms/lib # 'oracle' should not exist so delete it if it present % rm -f oracle % make -f ins_rdbms.mk no_parropt ioracle % exit NB: Do NOT deinstall the Parallel Server Option if the database is using Parallel Server unless both nodes are shut down otherwise database corruption could occur. A1.6) If the error is on STARTUP NOMOUNT: Check the init.ora file used to start the database. This provides the configuration details used configure the instance. To help isolate the problem it may be useful use a very basic init.ora file when starting the instance. If this works then parameters can be increased / introduced one at a time to see if there is a problem with a particular setting. A1.7) Check for server side trace files which may give more indication what the underlying problem is. See section C for details on how to check for server trace files. A1.8) Ensure there is free disk space in: a. Your USER_DUMP_DEST and BACKGROUND_DUMP_DEST locations b. Your AUDIT destination (Unix) The default is $ORACLE_HOME/rdbms/audit c. Your Oracle Trace directory if Oracle Trace is enabled See <Note:45482.1> A1.9) Windows 2000 - Only If the Server's sqlnet.ora file contains Authnetication services which are NOT reachable by Oracle, then an ORA-3113 error will result. For example, if the sqlnet.ora file contains the parameter: SQLNET.AUTHENTICATION_SERVICES = (NTS) and the Oracle database is moved from a Windows NT Domain to an Active Directory one, or if a Domain Controller is introduced, then an error will result trying to start the database. Remove the sqlnet.authentication_services line so that Oracle does not look for a non-existent KDC (Kerberos Domain Controller). A2) Errors Mounting the database Check all the items in A1 first. If an error occurs when mounting the database there may be problems with the control-files or data files, or with resources required to open these files. A2.1) The location of the control files are specified in the init.ora file. Try mounting using each controlfile in turn. eg: "Shutdown abort", edit the init.ora to refer to ONE of the controlfiles only, "startup nomount", "alter database mount" Repeat for each controlfile to see if any controlfile works. A2.2) It is possible to re-create the controlfiles if you know the location of all datafiles and online logs, or to restore an old backup controlfile. Always back up the current controlfiles before restoring any backup copies or issuing a CREATE CONTROLFILE command. The steps for this are not documented here. A2.3) Unix: Some unix platforms have a 'truss' command (or 'tusc'). If available this can be used to help trace how far Oracle gets before the error occurs. Eg: % truss -o /tmp/truss.out -f svrmgrl Keep the file /tmp/truss.out safe - Oracle Support MAY need to see it. A3) Errors on RECOVER DATABASE ORA-3113 during recover database is often related to a corruption on the database or redo stream which causes the shadow process to die. There should be a server side trace file produced for this sort of problem. See Section C for details on how to locate any trace files from both USER_DUMP_DEST and BACKGROUND_DUMP_DEST. A3.1) If the "recover database" fails fairly quickly then it may help to collect the redo up to the point of failure as this may help identify where the problem is. Use the following commands just prior to the RECOVER DATABASE command: SVRMGR> alter session set max_dump_file_size=unlimited; SVRMGR> alter session set events 2> '10228 trace name context forever, level 10'; SVRMGR> RECOVER DATABASE This causes redo information to be written to the user trace file. The last items of redo may help determine which file has problems. A3.2) If you do not have many datafiles in the database it may be as quick to recover each file in turn to see if this narrows down the problem. Eg: SVRMGR> select name from v$datafile; and then for each file: SVRMGR> RECOVER DATAFILE 'full_file_name' If this gets to a problem file then back up the file and use standard recovery options as if the file was lost. A4) Errors on ALTER DATABASE OPEN Database open performs very many operations and so it is necessary to collect any trace information before determining the next steps. However, the following may help isolate the problem more quickly: A4.1) Move files out of your USER_DUMP_DEST and BACKGROUND_DUMP_DEST directory as these steps will generate a lot of trace. A4.2) Edit the init.ora file and add the lines: event="10046 trace name context forever, level 12" event="10015 trace name context forever, level 1" event="10228 trace name context forever, level 1"